# Verify Code Functionality

Generated code can appear correct while still failing under real inputs. This skill guides Claude, Codex, or Claude Code through sandbox testing, debugging, and evidence-based reporting.

## Install

```bash
npx skillstore add dnyoussef/functionality-audit
```

## Metadata

- Status: approved
- Slug: dnyoussef-functionality-audit
- Skillstore revision: r1
- Version status: missing
- Tree hash: 8fd70736fe9a24606f0a469400a279e4b6d3a6a185aa71afbfacc28b7d62478a
- Author: DNYoussef
- GitHub username: DNYoussef
- License: MIT
- Repository: https://github.com/DNYoussef/ai-chrome-extension/tree/main/.claude/skills/functionality-audit
- Ref: 72d5025b022c77f7a51bdf5c1637c689c80e89d1
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Quality score: 79
- Quality tier: bronze
- Public page: https://skillstore.pages.dev/skills/dnyoussef-functionality-audit
- Manifest: https://skillstore.pages.dev/api/skills/dnyoussef-functionality-audit/manifest

## Capabilities

- Creates isolated test environments that avoid production side effects.
- Designs test cases for normal flows, boundaries, errors, and edge cases.
- Executes code with realistic inputs and verifies expected outputs.
- Analyzes failures through reproduction, tracing, and root cause investigation.
- Documents test results, identified bugs, remediation progress, and follow-up recommendations.

## Use Cases

- Validate Generated Code: Run focused tests after AI code generation to prove the implementation works with representative inputs.
- Debug Unclear Failures: Reproduce failures, inspect execution paths, and identify root causes before applying fixes.
- Prepare Release Checks: Perform final execution verification and document risks before shipping a change.

## Prompt Templates

### Run Basic Verification

```
Use the functionality-audit skill on these files. Confirm the code runs, test the main expected behavior, and report any failures with reproduction steps.
```

### Test Edge Cases

```
Audit this implementation with normal inputs, boundary values, malformed inputs, and empty inputs. Compare actual behavior against the expected behavior I provide.
```

### Debug a Failing Feature

```
Use functionality-audit to reproduce this failure, identify the first incorrect behavior, explain the root cause, and propose a minimal fix with regression tests.
```

### Verify a Release Candidate

```
Perform a full functionality audit for this release candidate. Create an isolated environment, run realistic tests, summarize pass and fail results, rank discovered bugs, and list remaining release risks.
```

## Limitations

- Requires clear expected behavior to judge whether outputs are correct.
- May need project-specific dependencies, test data, or services before execution.
- Cannot guarantee production behavior when important systems are mocked.
- Complex cloud or container environments may require extra setup time.

## Best Practices

- Provide expected behavior, representative inputs, and known edge cases before the audit starts.
- Run tests in an isolated environment that mirrors important production assumptions.
- Use audit results to add regression tests for every confirmed bug.

## Anti Patterns

- Assuming code works because it compiles or passes lint checks.
- Changing code before reproducing the failure and understanding the root cause.
- Testing only the easiest happy path while ignoring errors and boundaries.

## Security Audit

- Audited at: 2026-07-06T12:16:32.183\+00:00
- Summary: The static findings are false positives caused by prose about systematic testing and a DOT graph edge. I found no prompt injection, malware intent, data exfiltration, or instructions to perform unauthorized reconnaissance.

## Stats

- Views: 175
- Downloads: 7
- Favorites: 0
- Popularity score: 0